Provided is an autonomous vehicle including a storage configured to store a map including two-dimensionally represented road surface information and three-dimensionally represented structure information, a camera configured to obtain a two-dimensional (2D) image of a road surface in a vicinity of the vehicle, a light detection and ranging (LiDAR) unit configured to obtain three-dimensional (3D) spatial information regarding structures in a vicinity of the vehicle, and a controller comprising processing circuitry configured to determine at least one of the camera or the LiDAR unit as a position sensor, based on whether it is possible to obtain information regarding the road surface and/or the structures in the vicinity of the vehicle, to identify a position of the vehicle on the map corresponding to a current position of the vehicle using the position sensor, and performing autonomous driving based on the identified position on the map.
Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
1. An autonomous vehicle comprising: a storage configured to store a map including two-dimensionally represented road surface information and three-dimensionally represented structure information; a camera configured to obtain a two-dimensional (2D) image of a road surface in a vicinity of the vehicle; a light detection and ranging (LiDAR) unit comprising light detection and ranging circuitry configured to obtain three-dimensional (3D) spatial information regarding structures in the vicinity of the vehicle; and a controller comprising processing circuitry configured to: identify a position of the vehicle on the map corresponding to a current position of the vehicle by obtaining the 2D image of the road surface in the vicinity of the vehicle using the camera and mapping information regarding lanes and/or a road surface included in the 2D image to the two-dimensionally represented road surface information included in the map; determine the LiDAR unit as a position sensor based on determining that there are no lanes and/or road surface signs on a road surface on the map corresponding to the vicinity of the vehicle based on the two-dimensionally represented road surface information of the map; in response to the LiDAR unit being determined as the position sensor, identify the position of the vehicle on the map corresponding to the current position of the vehicle by obtaining the 3D spatial information regarding the structures in the vicinity of the vehicle using the LiDAR unit and mapping the 3D spatial information to the three-dimensionally represented structure information included in the map; and perform autonomous driving based on the identified position on the map.
2. The autonomous vehicle of claim 1 , wherein the controller is further configured to determine whether it is possible to obtain information regarding the road surface and/or the structures around the vehicle by determining whether there are lanes and/or road surface signs on a road surface in the vicinity of the vehicle on the map and/or whether there are structures in the vicinity of the vehicle on the map.
Autonomous vehicles rely on accurate environmental perception to navigate safely. A key challenge is determining whether sufficient information about the road surface and surrounding structures is available to support autonomous driving functions. This invention addresses this by using a controller that assesses the availability of relevant data from a map. The controller checks for the presence of lane markings or road surface signs on the map in the vicinity of the vehicle, indicating whether the road surface can be reliably identified. Additionally, the controller evaluates whether structures around the vehicle are mapped, ensuring that obstacles or reference points for navigation are accounted for. By cross-referencing the vehicle's location with map data, the system determines whether the necessary environmental information is present to enable safe and effective autonomous operation. This approach enhances situational awareness and decision-making for autonomous vehicles, particularly in dynamic or partially mapped environments.
3. The autonomous vehicle of claim 1 , wherein the controller is further configured to determine whether it is possible to obtain information regarding the road surface in the vicinity of the vehicle by determining whether it is possible to obtain information regarding lanes and/or road surface signs on the road surface from the 2D image of the road surface in the vicinity of the vehicle obtained by the camera.
This invention relates to autonomous vehicles equipped with systems for detecting and analyzing road surface conditions. The system includes a camera that captures a two-dimensional (2D) image of the road surface in the vicinity of the vehicle. A controller processes this image to determine whether it is possible to extract information about the road surface, specifically focusing on detecting lanes and road surface signs. The controller assesses the image to identify lane markings and other visual indicators that provide insights into road conditions, such as lane boundaries, road signs, or surface markings. If such information is detectable, the system uses it to enhance the vehicle's understanding of the road environment, improving navigation and safety. The invention addresses the challenge of accurately interpreting road conditions in real-time, ensuring that the autonomous vehicle can reliably navigate based on visual data. The system may also integrate with other sensors or data sources to supplement the 2D image analysis, ensuring robust performance under varying conditions.
4. The autonomous vehicle of claim 1 , wherein the controller is further configured to determine the LiDAR unit as a position sensor, based on determining that it is not possible to obtain information regarding lanes and/or road surface signs on the road surface from the 2D image.
Autonomous vehicles rely on various sensors to navigate and interpret road conditions. A key challenge is ensuring reliable detection of lane markings and road surface signs, which are critical for safe navigation. Traditional systems may struggle when visual sensors, such as cameras, fail to capture clear or sufficient data due to environmental factors like poor lighting, weather, or obstructions. This invention addresses this problem by enhancing the autonomous vehicle's sensor fusion capabilities. The vehicle includes a controller that dynamically selects the most reliable sensor for position tracking. If the camera fails to detect lane markings or road signs from a 2D image, the controller reassigns the LiDAR unit as the primary position sensor. LiDAR provides high-resolution 3D data, which can compensate for the camera's limitations by detecting lane structures and road features in three dimensions. This adaptive sensor selection improves navigation accuracy and reliability, especially in challenging conditions where visual sensors alone may be insufficient. The system ensures continuous and robust positioning data, enhancing the vehicle's ability to navigate safely and autonomously.
5. The autonomous vehicle of claim 1 , wherein the controller is further configured to determine whether structures capable of being sensed by the LiDAR unit are located in the vicinity of the vehicle, based on the map, and to determine the camera as a position sensor based on determining that there are no structures in the vicinity of the vehicle.
Autonomous vehicles rely on various sensors, including LiDAR and cameras, to navigate and perceive their environment. LiDAR is highly effective for detecting and mapping structures like buildings, trees, and other obstacles, but its performance can degrade in open areas with few or no detectable structures. This limitation can lead to reduced accuracy in vehicle positioning and navigation. To address this, an autonomous vehicle system is designed to dynamically select the most appropriate sensor for positioning based on environmental conditions. The system includes a LiDAR unit for detecting structures and a camera for capturing visual data. A controller accesses a map of the vehicle's surroundings to assess whether detectable structures are present in the vicinity. If the map indicates that no structures are available for LiDAR sensing, the controller designates the camera as the primary position sensor. This adaptive approach ensures reliable positioning even in open or sparsely structured environments, improving navigation accuracy and safety. The system may also include additional sensors and processing modules to enhance environmental awareness and decision-making.
6. The autonomous vehicle of claim 1 , wherein the controller is further configured to determine both the camera and the LiDAR unit as position sensors, based on it being possible to obtain both information regarding the road surface in the vicinity of the vehicle and information regarding the structures in the vicinity of the vehicle and based on a driving situation of the vehicle including at least one of changing lanes, turning, or making a U-turn.
This invention relates to autonomous vehicles equipped with multiple sensor systems, specifically cameras and LiDAR units, used for navigation and environmental perception. The problem addressed is the need for reliable position sensing during complex driving maneuvers, such as changing lanes, turning, or making U-turns, where traditional sensor configurations may fail to provide sufficient data for accurate localization and obstacle detection. The autonomous vehicle includes a controller that dynamically selects and utilizes both camera and LiDAR sensors as position sensors under specific driving conditions. The controller determines that both sensors should be used when they can simultaneously provide information about the road surface and surrounding structures in the vicinity of the vehicle. This dual-sensor approach enhances situational awareness by combining visual data from the camera with high-resolution spatial data from the LiDAR unit, ensuring robust localization and obstacle detection during maneuvers that require precise positioning. The system is particularly useful in scenarios where single-sensor reliance may be insufficient, such as when road markings are obscured or when structural landmarks are needed for navigation. By leveraging both sensors, the vehicle can maintain accurate positioning and safely execute complex driving tasks. The invention improves the reliability and safety of autonomous driving by dynamically adapting sensor usage based on real-time driving conditions.
7. The autonomous vehicle of claim 1 , wherein the map comprises a map including information regarding a reliability index representing a degree of reliability of each object in the map, and the controller is further configured to identify the position of the vehicle on the map corresponding to the current position of the vehicle based on the reliability index of each object in the map.
This invention relates to autonomous vehicle navigation systems that improve localization accuracy by incorporating reliability indices for map objects. The system addresses the challenge of accurately determining a vehicle's position in dynamic or uncertain environments where map data may be incomplete or unreliable. The autonomous vehicle includes a map database that stores a reliability index for each object, representing the confidence level in the object's accuracy or relevance. A controller uses this reliability index to weigh the importance of each map object when determining the vehicle's position. By prioritizing highly reliable objects, the system enhances localization precision, reducing errors caused by outdated or inaccurate map data. The reliability index may be dynamically updated based on sensor data, historical accuracy, or environmental conditions. This approach ensures the vehicle's position is calculated using the most trustworthy map information available, improving navigation safety and efficiency in real-world scenarios. The system is particularly useful in environments where map data may vary in quality, such as urban areas with frequent construction or rural regions with sparse mapping.
8. An autonomous driving method comprising: storing a map including two-dimensionally represented road surface information and three-dimensionally represented structure information; identifying a position of a vehicle on the map corresponding to a current position of the vehicle by obtaining a two-dimensional (2D) image of the road surface in vicinity of the vehicle using a camera and mapping information regarding lanes and/or a road surface included in the 2D image to the two-dimensionally represented road surface information included in the map; determining a LiDAR unit as a position sensor based on determining that there are no lanes and/or road surface signs on a road surface on the map corresponding to the vicinity of the vehicle based on the two-dimensionally represented road surface information of the map; in response to the LiDAR unit being determined as the position sensor, identifying the position of the vehicle on the map corresponding to the current position of the vehicle by obtaining three-dimensional (3D) spatial information regarding the structures in the vicinity of the vehicle using the LiDAR unit and mapping the 3D spatial information to the three-dimensionally represented structure information included in the map; and performing autonomous driving based on the identified position on the map.
The invention relates to autonomous driving systems that adapt their positioning methods based on available road surface and structural information. The system uses a map containing both two-dimensional (2D) road surface data (e.g., lanes, road markings) and three-dimensional (3D) structural data (e.g., buildings, landmarks). For positioning, the system first captures a 2D image of the road surface near the vehicle using a camera and matches lane or road surface features from the image to the 2D map data to determine the vehicle's position. If the map indicates no lanes or road surface signs in the vicinity, the system switches to using a LiDAR unit as the primary sensor. The LiDAR captures 3D spatial data of nearby structures, which is then matched to the 3D map data to localize the vehicle. Autonomous driving is executed based on the determined position. This approach ensures reliable localization even in areas with insufficient road surface markings by leveraging 3D structural data when necessary.
9. The autonomous driving method of claim 8 , further comprising determining whether it is possible to obtain the information regarding the road surface and/or the structures in the vicinity of the vehicle by determining whether there are lanes and/or road surface signs on a road surface in the vicinity of the vehicle on the map and/or whether there are structures in the vicinity of the vehicle on the map.
This invention relates to autonomous driving systems, specifically methods for assessing the availability of road surface and structural information in a vehicle's vicinity. The technology addresses the challenge of ensuring that autonomous vehicles have sufficient environmental data to navigate safely and accurately. The method involves determining whether relevant information about the road surface (such as lanes or road surface signs) or nearby structures is available. This is done by checking a map to see if lanes or road surface signs are present on the road near the vehicle or if structures are present in the vehicle's vicinity. If such information is available, the system can use it to enhance navigation and decision-making. If not, the system may rely on alternative data sources or adjust its operations accordingly. The method ensures that the autonomous vehicle has the necessary data to operate effectively, improving safety and reliability in various driving conditions.
10. The autonomous driving method of claim 8 , further comprising determining whether it is possible to obtain the information regarding the road surface in the vicinity of the vehicle by determining whether it is possible to obtain information regarding lanes and/or road surface signs on the road surface from the 2D image of the road surface in the vicinity of the vehicle obtained by the camera.
This invention relates to autonomous driving systems that analyze road surface conditions using camera-based imaging. The problem addressed is the need for reliable detection of road surface information, such as lane markings and road signs, to support safe and accurate autonomous navigation. Existing systems may struggle with low visibility, occlusions, or environmental factors that obscure such details. The method involves capturing a 2D image of the road surface near the vehicle using a camera. The system then processes this image to determine whether it can extract relevant information about lanes and road surface signs. If the image data is sufficient, the system proceeds to obtain and utilize this information for navigation. If not, alternative methods or data sources may be employed. The approach ensures that the autonomous vehicle can adaptively assess the availability of critical road surface details before relying on them for decision-making. This enhances robustness in varying driving conditions, reducing the risk of errors due to incomplete or unreliable visual data. The technique is particularly useful in dynamic environments where road markings or signs may be partially obscured or degraded.
11. The autonomous driving method of claim 8 , further comprising determining the LiDAR unit as a position sensor, based on it being determined that there are no lanes and/or road surface signs on a road surface on the map corresponding to the vicinity of the vehicle.
This invention relates to autonomous driving systems, specifically methods for dynamically selecting sensor configurations based on environmental conditions. The problem addressed is the challenge of accurately determining vehicle position in scenarios where traditional road markings or signs are absent, such as in rural areas or poorly maintained roads. The solution involves using a LiDAR unit as a position sensor when the system detects that no lanes or road surface signs are present in the map data corresponding to the vehicle's vicinity. The LiDAR unit generates three-dimensional point cloud data of the surrounding environment, which is then processed to estimate the vehicle's position relative to detected objects or terrain features. This adaptive approach ensures reliable localization even in the absence of conventional road infrastructure. The method may also involve integrating data from other sensors, such as cameras or radar, to enhance position accuracy. The system dynamically adjusts sensor usage based on real-time environmental analysis, improving robustness in diverse driving conditions.
12. The autonomous driving method of claim 8 , further comprising determining the LiDAR unit as the position sensor, based on it being determined that obtaining information regarding lanes and/or road surface signs on the road surface from the 2D image is not possible.
This invention relates to autonomous driving systems, specifically methods for selecting position sensors when visual data from cameras is insufficient. The problem addressed is the inability to reliably detect lanes or road surface signs from 2D images due to environmental conditions like poor lighting, occlusion, or adverse weather. In such cases, the system automatically switches to using a LiDAR unit as the primary position sensor. The LiDAR unit provides 3D spatial data, which is more robust in challenging conditions where 2D image processing fails. The method involves analyzing the 2D image data to determine if lane or road sign information can be extracted. If not, the system defaults to the LiDAR unit for position sensing, ensuring continuous and reliable navigation. This approach improves the reliability of autonomous driving by dynamically adapting sensor usage based on environmental conditions. The system may also integrate other sensors, such as radar or inertial measurement units, to enhance accuracy and redundancy. The method ensures that the vehicle maintains safe and accurate positioning even when visual data is compromised.
13. The autonomous driving method of claim 8 , further comprising determining whether structures capable of being sensed by the LiDAR unit are located in the vicinity of the vehicle, based on the map, and determining the camera as a position sensor based on it being determined that there are no structures in the vicinity of the vehicle.
This invention relates to autonomous driving systems, specifically improving sensor selection for vehicle positioning in environments where traditional sensors like LiDAR may be ineffective. The problem addressed is the inability of LiDAR to reliably detect positioning references in certain environments, such as open areas with few or no detectable structures. The solution involves dynamically selecting alternative sensors, such as cameras, when LiDAR cannot provide sufficient data for accurate vehicle positioning. The method first accesses a map of the vehicle's surroundings to identify whether structures capable of being sensed by the LiDAR unit are present in the vicinity. If no such structures are detected, the system determines that the camera should be used as the primary position sensor. This adaptive approach ensures reliable positioning even in environments where LiDAR performance is compromised, enhancing the robustness of autonomous driving systems. The method may also include generating a positioning signal based on data from the selected sensor, ensuring continuous and accurate vehicle localization. This adaptive sensor selection improves autonomous vehicle operation in diverse environments, particularly where traditional LiDAR-based positioning may fail.
14. The autonomous driving method of claim 8 , further comprising determining both the camera and the LiDAR unit as position sensors, based on it being possible to obtain both information regarding the road surface in the vicinity of the vehicle and information regarding the structures in the vicinity of the vehicle and based on a driving situation of the vehicle including at least one of changing lanes, turning, or making a U-turn.
This invention relates to autonomous driving systems that utilize both camera and LiDAR sensors to enhance vehicle positioning accuracy during specific driving maneuvers. The problem addressed is the need for precise localization in dynamic driving scenarios where traditional sensor data may be insufficient. The method involves determining when both the camera and LiDAR unit should function as position sensors, which occurs when the system can simultaneously obtain information about the road surface and nearby structures. This dual-sensor approach is particularly useful during maneuvers such as lane changes, turns, or U-turns, where accurate positioning is critical. The camera captures visual data of the road and surroundings, while the LiDAR provides high-resolution depth and structural information. By combining these inputs, the system improves localization accuracy, ensuring safer and more reliable autonomous driving in complex scenarios. The method dynamically adjusts sensor usage based on real-time driving conditions, optimizing performance without relying on additional hardware. This approach enhances situational awareness and reduces dependency on a single sensor type, addressing limitations in traditional autonomous navigation systems.
15. The autonomous driving method of claim 8 , wherein the map comprises a map including information regarding a reliability index representing a degree of reliability of each object in the map, and the method further comprises identifying the position of the vehicle on the map corresponding to the current position of the vehicle based on the reliability index of each object in the map.
This invention relates to autonomous driving systems, specifically improving vehicle localization accuracy by leveraging a map with reliability indices for map objects. The problem addressed is the challenge of precisely determining a vehicle's position in dynamic environments where map data may contain uncertainties or outdated information. The method involves using a map that includes reliability indices for each object, where the reliability index quantifies the confidence level of the object's accuracy or relevance. The system identifies the vehicle's position on the map by comparing the current vehicle position with the map objects, weighted by their reliability indices. This ensures that more reliable map features contribute more significantly to localization, reducing errors caused by unreliable or obsolete map data. The approach enhances localization by dynamically adjusting the influence of map objects based on their reliability, improving navigation in environments with varying data quality. This method is particularly useful in scenarios where map data may be incomplete or subject to frequent changes, such as urban areas with construction or temporary obstacles. The system dynamically updates the vehicle's position by prioritizing high-reliability map features, ensuring more accurate and robust autonomous driving.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
April 29, 2019
February 22, 2022
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.